*3.2.2. Phylogenetic analysis based on SLAM and morbillivirus H proteins*

188 New Approaches to the Study of Marine Mammals

viral release into the air.

**3.2. Structure of SLAMs of marine mammals** 

sequences to compare the interface of their SLAM V domains [71].

*3.2.1. Primary structure of SLAM proteins* 

that of the elephant (86%) [71].

showed that MV initially targets SLAM-positive immune cells such as alveolar macrophages, dendritic cells, and lymphocytes, and later the viral infection spreads to the epithelial cells of the trachea, lungs, oral cavity, pharynx, or intestines, which are SLAMnegative cells [66,67]. Another infection experiment using epithelial cell receptor-blind MV, demonstrated that the mutant MV inoculated intranasally to monkeys shows virulence and infectivity toward lymphoid tissues, although the virus cannot cross the airway epithelium and cannot be shed in the air [68]. The molecule forms tight junctions on polarized epithelial cells and was shown to function as the receptor for effectively releasing MV to the apical side of epithelial cells [69,70]. This explains why MV is highly contagious. Thus, the wildtype MV posesses two types of receptor, SLAM for entry and propagation and Nectin 4 for

Cetaceans and sirenians have achieved complete adaptation to the aquatic environment and spend all of their lives in water. Cetaceans belong to the order Cetartiodactyla, superorder Laurasiatheria, and are closely related to hippopotami or ruminants among land animals. Sirenians, including dugongs and manatees, are in the order Sirenia, superorder Afrotheria, and are evolutionarily related to elephants or hyraxes. Pinnipeds, belonging to the order Carnivora, superorder Laurasiatheria, are not completely adapted to the aquatic environment and they must deliver and nurse their young on land. This characteristic of pinnipeds makes it possible to transmit infectious diseases between aquatic and land mammals. To determine the structure of marine mammal SLAMs, we collected blood samples from taxonomically different animal groups, i.e., cetaceans, pinnipeds, and sirenians. White blood cells were obtained from: two species of cetacean, a Pacific white-sided dolphin (*Lagenorhynchus obliquidens*) and a killer whale (*Orcinus orca*); two species of pinniped, a spotted seal (*Phoca largha*), and a walrus (*Odobenus rosmarus*); and a sirenian, a West Indian manatee (*Trichechus manatus*). The blood of an Indian elephant (*Elephas maximus bengalensis*) was also collected. After immune stimulation with phytohemagglutinin, RNAs of the leukocytes were extracted. First, the complete nucleotide sequences of the SLAM genes were determined. Three-dimensional models were then generated based on the deduced amino acid

The deduced amino acid sequences of marine mammal and elephant SLAMs indicated that they contain 336–339 amino acid residues, inducing six cysteine residues and six potential *N*-linked glycosylation sites (Figure 1). They have two immunoglobulin-like domains, V and C2, in the extracellular region, and two ITSM motifs (T-X-Y-X-X-V/I) and one ITSM-like sequence in the intracellular region. These molecular features are shared with all reported mammalian SLAMs (Figure 1). The cetacean and pinniped SLAMs showed the greatest homology with those of artiodactyla (cow and sheep, 84–85% identity at the amino acid level) and of dogs (84%), respectively. Manatee SLAM shared the greatest homology with Phylogenetic trees based on SLAM and the morbillivirus H protein were constructed using the maximum-likelihood (ML) and Bayesian methods. In the phylogeny of SLAMs (Figure 2(a)), each taxonomic group, including primates (humans, chimpanzees, rhesus monkeys, and marmosets), cetaceans (Pacific white-sided dolphins and killer whales), artiodactyls (cows, buffalo, sheep, and goats), pinnipeds (spotted seals and walruses), and rodents (mice and rats), was monophyletic with a 100% ML bootstrap probability (BP) and a 1.00 Bayesian posterior probability (BPP). Manatee and elephant SLAMs, dog and pinipped SLAMs, and cetacean and artiodactyl SLAMs formed single clades, each with a 100% BP and a 1.00 BPP value, respectively.

Morbillivirus phylogeny based on MV H protein sequences reflected the host grouping, except for MV. CDV (dogs, Baikal seals) and PDV (seals), and PMV (porpoises) and DMV (dolphins), respectively, formed single clades each with 100% BP and 1.00 BPP support (Figure 2(b)). The monophyletic lineage of MV (human) (100% BP and 1.00 BPP) was within the grouping of ruminant viruses, RPV (cow) and PPRV (sheep and goat), with 100% BP and 1.00 BPP. These phylogenetic trees indicated that SLAMs and viral H proteins roughly coevolved. However, the monophyletic lineage of MV and the ruminant viruses RPV and PPRV suggested that human MV may have originated from ancestral RPV in cattle by acquiring a binding affinity for human SLAM, as proposed in a previous report based on the morbillivirus P gene [9]

Host-Virus Specificity of the Morbillivirus Receptor, SLAM, in Marine Mammals:

crystal structure information of the bound complex of MV H and marmoset SLAM V [56]. Figure 3 shows the 3D model of the Pacific white-sided dolphin SLAM extracellular domain in a self-ligand form. In the models, the V and C2 domains are shown with rod-like structures, which are both constituted mainly of -sheets containing several -strands. The cysteine residues appear to be important in forming the basic 3D structures. On the basis of amino acid sequence similarity, this structure is shared with all of the SLAMs examined. The V domain possesses a two-layered -sheet structure, and the front sheets provide an

In (a), the blue and green models show respective SLAM extracellular domains forming a homophilic dimer. The βstrands are indicated by arrows and the disulfide bonds are shown as yellow bars. The thick red arrow indicates the direction of view of the front face of the morbillivirus binding site, as shown in Figure 4. In (b), the blue and black models indicate two SLAM molecules of two adjacent cells, respectively. The Protein Data Bank (PDB) entries for the marmoset SLAM in the complex (3ALW: A-D chains, 3ALX: A–D chains, 3ALZ: B chain; reference [56]) and for human NTB-A (21F7: A–D chains; reference [75]), were used as the template structure. The 3D model was constructed using the MODELLER 9.10 program [76] and visualized using PyMOL 1.4.1 (Schrodinger LLC) and PovRay (Persistence of

**Figure 3.** Ribbon diagram of the 3D structure (a) and schematic drawing (b) of the SLAM extracellular

Figure 4 shows top views of the front face of the modeled V domains of SLAMs of the spotted seal, Pacific white-sided dolphin, and West Indian manatee. The amino acid residues that have protruding side-chains on their front faces are likely a component of virus binding. We found such qualified 27 amino acid residues; 12 amino acid residues on the β-strands and 15 residues on loops (Figure 4). In addition, the amino acid residues positioned at 76, and the residues at 127-131 are thought to be important for the binding of

interface for binding with morbilliviruses.

Vision Pty. Ltd.).

domain from the Pacific white-sided dolphin.

Risk Assessment of Infection Based on Three-Dimensional Models 191

ML analyses were performed using PhyML [72], and an input tree was generated using BIONJ with the JTT model [73] along with amino acid substitution that incorporated invariable sites and used a discrete gamma distribution (eight categories) (JTT + I + G model). Bayesian phylogenetic analyses were conducted using MrBayes version 3.0 within the JTT + I + G model [74]. The ML bootstrap probabilities and BPPs are shown at the nodes. Host animals and the corresponding morbilliviruses are shown as circles with the same alphabetic notations: A, cetaceans; B, artiodactyla; C, pinnipeds; D, dogs; E, primates; F, rodents; and G, manatees and elephants. Morbillivirus has not been identified in rodents (F), or manatees and elephants (G). The following SLAM gene sequences were obtained from the Data Bank of Japan: Pacific white-sided dolphin (AB428366); killer whale (AB428367); spotted seal (AB428368); walrus (AB428369); Indian elephant (AB428370); American manatee (AB428371); human beings (*Homo sapiens,* NM\_003037); chimpanzee (*Pan troglodytes*, XM\_513924); marmoset (*Saguinus oedipus*, AF257239); Rhesus monkey (*Macaca mulatta*, XM\_001117605); cow (*Bos taurus*, AF329970); buffalo (*Bubalus bubalis*, DQ228868); sheep (*Ovis aries*, NM\_001040288); goat (*Capra hircus*, DQ228869); dog (*Canis familaris*, AF325357); mouse (*Mus musculus*, NM\_013730); and rat (*Rattus norvegicus*, XM\_001054873). Morbillivirus H protein sequences were obtained as follows: MV Edmonston AIK-C strain (AB046218); MV V9301 strain (AB012948); MV B9301 strain (AB012949); RPV RBOK strain (Z30697); PPRV Turkey 2000 strain (NC\_006383); CDV Onderstepoort vaccine strain (AF305419); CDV Baikal seal strain (X84998); CDV 5804/Han90 strain (X85000); PDV Ulster/88 strain (D10371); PMV 2990 strain (AY586537); and DMV (NC\_005283). The alignments of the deduced amino acid sequences from these genes were generated using ClustalW version 1.8, inspected visually, and edited manually. This figure was adapted from [71].

**Figure 2.** Phylogenetic trees of SLAM peptide sequences (a) and morbillivirus H proteins (b).

#### *3.2.3. Three-dimensional models of marine mammal SLAM extracellular domain*

In order to analyze the binding site for morbillivirus, three-dimensional (3D) models of marine mammal SLAM extracellular domains were generated by homology modeling. Previously, we constructed 3D homology models based on the crystallographic structure of the human NTB-A molecule, a member of the SLAM family, as a template [71,75]. In the present study, we generated a new version of the models by adding the recently determined crystal structure information of the bound complex of MV H and marmoset SLAM V [56]. Figure 3 shows the 3D model of the Pacific white-sided dolphin SLAM extracellular domain in a self-ligand form. In the models, the V and C2 domains are shown with rod-like structures, which are both constituted mainly of -sheets containing several -strands. The cysteine residues appear to be important in forming the basic 3D structures. On the basis of amino acid sequence similarity, this structure is shared with all of the SLAMs examined. The V domain possesses a two-layered -sheet structure, and the front sheets provide an interface for binding with morbilliviruses.

190 New Approaches to the Study of Marine Mammals

ML analyses were performed using PhyML [72], and an input tree was generated using BIONJ with the JTT model [73] along with amino acid substitution that incorporated invariable sites and used a discrete gamma distribution (eight categories) (JTT + I + G model). Bayesian phylogenetic analyses were conducted using MrBayes version 3.0 within the JTT + I + G model [74]. The ML bootstrap probabilities and BPPs are shown at the nodes. Host animals and the corresponding morbilliviruses are shown as circles with the same alphabetic notations: A, cetaceans; B, artiodactyla; C, pinnipeds; D, dogs; E, primates; F, rodents; and G, manatees and elephants. Morbillivirus has not been identified in rodents (F), or manatees and elephants (G). The following SLAM gene sequences were obtained from the Data Bank of Japan: Pacific white-sided dolphin (AB428366); killer whale (AB428367); spotted seal (AB428368); walrus (AB428369); Indian elephant (AB428370); American manatee (AB428371); human beings (*Homo sapiens,* NM\_003037); chimpanzee

(*Pan troglodytes*, XM\_513924); marmoset (*Saguinus oedipus*, AF257239); Rhesus monkey (*Macaca mulatta*,

**Figure 2.** Phylogenetic trees of SLAM peptide sequences (a) and morbillivirus H proteins (b).

*3.2.3. Three-dimensional models of marine mammal SLAM extracellular domain* 

In order to analyze the binding site for morbillivirus, three-dimensional (3D) models of marine mammal SLAM extracellular domains were generated by homology modeling. Previously, we constructed 3D homology models based on the crystallographic structure of the human NTB-A molecule, a member of the SLAM family, as a template [71,75]. In the present study, we generated a new version of the models by adding the recently determined

and edited manually. This figure was adapted from [71].

XM\_001117605); cow (*Bos taurus*, AF329970); buffalo (*Bubalus bubalis*, DQ228868); sheep (*Ovis aries*, NM\_001040288); goat (*Capra hircus*, DQ228869); dog (*Canis familaris*, AF325357); mouse (*Mus musculus*, NM\_013730); and rat (*Rattus norvegicus*, XM\_001054873). Morbillivirus H protein sequences were obtained as follows: MV Edmonston AIK-C strain (AB046218); MV V9301 strain (AB012948); MV B9301 strain (AB012949); RPV RBOK strain (Z30697); PPRV Turkey 2000 strain (NC\_006383); CDV Onderstepoort vaccine strain (AF305419); CDV Baikal seal strain (X84998); CDV 5804/Han90 strain (X85000); PDV Ulster/88 strain (D10371); PMV 2990 strain (AY586537); and DMV (NC\_005283). The alignments of the deduced amino acid sequences from these genes were generated using ClustalW version 1.8, inspected visually,

In (a), the blue and green models show respective SLAM extracellular domains forming a homophilic dimer. The βstrands are indicated by arrows and the disulfide bonds are shown as yellow bars. The thick red arrow indicates the direction of view of the front face of the morbillivirus binding site, as shown in Figure 4. In (b), the blue and black models indicate two SLAM molecules of two adjacent cells, respectively. The Protein Data Bank (PDB) entries for the marmoset SLAM in the complex (3ALW: A-D chains, 3ALX: A–D chains, 3ALZ: B chain; reference [56]) and for human NTB-A (21F7: A–D chains; reference [75]), were used as the template structure. The 3D model was constructed using the MODELLER 9.10 program [76] and visualized using PyMOL 1.4.1 (Schrodinger LLC) and PovRay (Persistence of Vision Pty. Ltd.).

**Figure 3.** Ribbon diagram of the 3D structure (a) and schematic drawing (b) of the SLAM extracellular domain from the Pacific white-sided dolphin.

Figure 4 shows top views of the front face of the modeled V domains of SLAMs of the spotted seal, Pacific white-sided dolphin, and West Indian manatee. The amino acid residues that have protruding side-chains on their front faces are likely a component of virus binding. We found such qualified 27 amino acid residues; 12 amino acid residues on the β-strands and 15 residues on loops (Figure 4). In addition, the amino acid residues positioned at 76, and the residues at 127-131 are thought to be important for the binding of

the virus, because the side-chains of these residues are closely located to those of viral H protein in a crystal structure of the complex [56]. Particularly, the residues at 127-131 are thought to form an intramolecular β-sheet with the β-strand of MV H [56]. The overall 3D structures of the interfaces are similar among SLAMs, but several among the total 32 amino acid residues possibly contributing the binding affinity to the virus, differed among the three marine mammals.

Host-Virus Specificity of the Morbillivirus Receptor, SLAM, in Marine Mammals:

*3.2.4. Prediction of amino acid residues involved in virus binding and host–virus* 

different between dolphin and seal, and between dolphin and humans, respectively.

58 K K K K K K K 60 I I I I I I I 61 H H H H H H R\* 63 L L L L L V V 65 T T T T T T T 67 A A A A A A E\* 68 E E G\* E E K\* T\* 69 S S S S S S S 72 N N D D D N S 73 S S T T T S T 74 V I V V V V F 75 K K K K K E\* K 76 K K K K K N\* K 77 K K K K K K K 80 S S S S S S S 82 D D D D D D D 84 P P R R R S S 85 E E K K K E E 87 G G D D G G G 90 R R H R H R P\* 92 L L L L L L L 117 W W W W W W W 119 F F F F F L F 121 T T S S S T T 123 E E E E E E E

a.a. No. Seal Dog Dolphin Cow Sheep Human Manatee

To identify amino acid residues that are important for host–virus specificity, we compared the 32 residues with those of land mammal SLAMs (Table 2). The difference in the SLAM interface was only two amino acid residues between seal and dog (Val and Ile at position 74, and Arg and Gln at position 129), between human and marmoset (Leu and Phe at position 119, Val and Ile at position 126, marmoset data not shown), and between cow and sheep (Asp and Gly at position 87, and Arg and His at position 90). This is consistent with the evidence that mass die-offs of Baikal seals and Caspian seals were caused by CDV; marmosets are highly sensitive to MV; and that RPV and PPRV can infect ruminants. It is noted that the identity of these 32 residues between dolphin and cow SLAMs is very high, although they are infected by different morbilliviruses, CMV and RPV, respectively. Four residues are different between the two animals, while eleven and fourteen residues are

*specificity* 

Risk Assessment of Infection Based on Three-Dimensional Models 193

The interfaces are viewed from the direction shown by the thick red arrow in Figure 3. The amino acid residues that possibly interact with the viral H protein, are shown with their position numbers. The different amino acid residues among the three mammal SLAMs are indicated in black, and their side-chains are shown with the atoms colored (black for carbons, blue for nitrogens, and red for oxygens). The shared residues among the three are shown in blue. Disulfide bonds are shown in yellow.

**Figure 4.** Ribbon diagram of the 3D structure models of the SLAM interface for binding morbillivirus from the spotted seal (a), Pacific white-sided dolphin (b), and manatee (c).
